On February 24/25 2015, the peace informatics lab and the Dutch ministry of foreign affairs joined forces to organise a two-days meeting around responsible data for humanitarian response in Den Haag. For a start I must say the ministry of foreign affairs has extremely nice meeting facilities and a building so complex it makes it really worth having someone accompany you at all time! 😉 It was also interesting that the organisers had decided to put strict confidentiality rules such as doing no reporting of the discussions during the break out sessions and being cautious with taking pictures during the plenary. Being used to academic conferences and open data events where everything is always very shared and public this came as an intriguing, though not frustrating, surprise.
The event was organised around 3 themes : 1) data deficit, 2) human-centred design and 3) data governance. Here are some of the things that were being discussed:
- Big data is both a great source of opportunities and risks. Without data it is hard to provide a swift, targeted and efficient response to crisis scenarios but without care data driven process go out of control. It is important to keep the focus on the people that need help, to engage with them and avoid blind decision processes enforcing de-contextualised solutions to their needs. The case could further be made that a regulator should be put in place to ensure whose willing to help adhere to responsible data practices.
- It is easy to think big data is everywhere but in fact a split can be made between data poor and data rich environments. Doing doing collection, analyse and dispatch of data in data poor environments is a challenge to be tackled. This could be facilitated by having more standards for sharing data, make the value of data explicit, and encourage local civic organisations to produce data feeds.
- Humanitarian response is a much self-organised and local process. There are also different dynamics to be observed when looking at the people and group level. Crowds are also known to be the most innovative entity, not the individuals on their own. When working together people get more efficient at problem solving. As a system, communities exhibit complex systems behaviour and have to be considered as such. Moving from isolated datasets and disciplines to multi-disciplinary approaches tackling interconnected datasets require us to re-consider how we approach development and humanitarian challenges.
- In the end data-driven aid and services remains about the people it is aimed to help. It is not about the data itself though it has to be treated with a lot of care. To understand the needs of people it is best to stay with them for a while rather than asking them what they want, and even worse asking that remotely.
- When asking around everyone will agree human-centred design is a good thing to do and claim they do it in some way or another. In practice the situation can be different as some may take “having users” as being human-centred design, but the same practice of putting users at the centre of the design of systems can also be found under different names. Whatever what is done and how it is called good ethics, safeguards, and community engagement are essentials things to keep an eye on.
- Big data is commonly used to understand things, find patterns and doing predictive analysis. This later point is very much at risk: in, say, an immigration context knowing that a given amount of people is likely to cross a particular country border at one point can be used to anticipate on a matching action. Depending on the context and point of views this action can be supportive of the flow (provision of food, shelters, …) or repressive against it (closed doors, guns, …). Other predictions can be used to either supporting social minorities/groups or oppressing them. Then raise some challenging issues: If you make predictive analysis, who do you share the outcomes with ? Under which conditions ? Is accountability different in cases where everyone can make the same predictions using open algorithms and open data ?
- While anonymity of persons is known well understood and done there is a blatant need for practices and understanding of the anonymity of groups. Human-centred design should also take care of upper levels of data aggregation (communities, organisations, countries, …).
- It was not put as such by anyone but it can be said, or better repeated, that, unlike humans, not all datasets are born equal. There are datasets that can surely be open, some that can surely not be open, and a wide spectrum of different levels of access between these two extremes. There are also datasets that are more complete than others as sampling and data collection techniques can lead to forgetting the “digital invisibles” among the population. It is important to understand (when using) and describe (when publishing), at least, what a dataset exactly contains, how sensitive the information is, how it was generated and how it can be used (see the FIPs for a more complete list).
- Some countries have no legal framework for data. The correlation between power and data is nonetheless an universal constant. Humanitarianism is a driving force nobody can really be against. The combination of these three facts can lead to observing, without much range for actions, uncontrolled data-driven bad practices operating under cover of doing good things.
- All this should nonetheless no make us confuse fear and risk. Inertia is the biggest threat here. It is important to think things through and make use of professional audits for security and legal aspects in order to make clear and informed choices.
Some more take-home things from the event:
- The Data for Climate Action (D4CA) announced at the end of the event is a very interesting other incarnation of the interest Orange has into using his data for tackling major challenges. This, in a way, follows upon two editions of Data for Development (D4D) and, I hope, precedes new upcoming “Data for” initiatives.
- Data for humanitarian response, Data for development and Data for the humanities share very similar concerns about doing things “responsibly” (possibly with some slightly different understanding of the meaning of that word). It would be interesting to look into the overlap among their respective practices, guidelines and legal frameworks.
- The living labs approach we use in our projects is most probably a good way to do human-centred design but we could, and should, spend more time on ethical and privacy issues around the data.
- The break-out sessions where very neatly organised and lead to lively discussions. Asking everyone to filter out the post-its to highlights potential immediate and targeted actions is a great idea!
As a closing remark, I would like to say that this was a great event which I attended with pleasure and interest but I missed a bit on the practical side. Having more case-based evidences and discussions would have been a plus. Great inspiring talks are a very good thing for motivation but they do not correlate much (IMHO) with fostering new ideas, on the contrary looking at what each other has been doing and getting into collaborative complex problem solving in “hands-on” sessions can lead to many new and unexpected ideas. Especially when considering such a diverse crowd.